Band Tailing and Deep Defect States in CH3NH3Pb(I1−xBrx)3 Perovskites As Revealed by Sub-Bandgap Photocurrent

Organometal halide perovskite semiconductors have emerged as promising candidates for optoelectronic applications because of the outstanding charge carrier transport properties, achieved with low-temperature synthesis. Here, we present highly sensitive sub-bandgap external quantum efficiency (EQE) measurements of Au/spiro-OMeTAD/CH3NH3Pb(I1−xBrx)3/TiO2/FTO/glass photovoltaic devices. The room-temperature spectra show exponential band tails with a sharp onset characterized by low Urbach energies (Eu) over the full halide composition space. The Urbach energies are 15−23 meV, lower than those for most semiconductors with similar bandgaps (especially with Eg > 1.9 eV). Intentional aging of CH3NH3Pb(I1−xBrx)3 for up to 2300 h, reveals no change in Eu, despite the appearance of the PbI2 phase due to decomposition, and confirms a high degree of crystal ordering. Moreover, sub-bandgap EQE measurements reveal an extended band of sub-bandgap electronic states that can be fit with one or two point defects for pure CH3NH3PbI3 or mixed CH3NH3Pb(I1−xBrx)3 compositions, respectively. The study provides experimental evidence of defect states close to the midgap that could impact photocarrier recombination and energy conversion efficiency in higher bandgap CH3NH3Pb(I1−xBrx)3 alloys. Organometal halide perovskites, with the general formula ABX3, have generated enormous interest over the last 7 years as their solar cell power conversion efficiencies have increased from 3.8% in 2009 to the current world record of 22.1%. Much of this progress has been realized by composition engineering at the A site with cations including CH(NH2)2 , CH3NH3 , and Cs, at the B site with Pb and Sn cations, and at the X site with I−, Br−, and Cl− anions. These changes have allowed researchers to increase photovoltaic power conversion efficiencies and chemical stabilities, as well as engineer the bandgap (Eg) for use in high-efficiency tandem photovoltaic devices. Despite the intense interest in composition engineering, the effects of alloying on structural disorder and the densities of states (DOS), particularly within the bandgap, are not well documented. Their characterization is of central importance to identifying factors that contribute to efficiency loss and instability for a number of reasons. First, defects cause nonradiative recombination and modify the band alignment. Second, there is a growing consensus that current−voltage hysteresis and photocurrent transients in organometal perovskite solar cells are due to defect states and related mobile ionic species. Third, alloying might be expected to increase disorder in a crystal as the variety of possible defect states is increased. Importantly, the optical band edge, characterized by the Urbach energy (Eu), 9 can be broadened by disorder in the crystal lattice (or by impurities). Low Urbach energies are highly desirable for any semiconductor used in optoelectronic devices. Finally, the evolution of defect states and structural disorder upon aging could play a role in cell degradation. Received: December 23, 2016 Accepted: February 15, 2017 Published: February 15, 2017 Leter http://pubs.acs.org/journal/aelccp © 2017 American Chemical Society 709 DOI: 10.1021/acsenergylett.6b00727 ACS Energy Lett. 2017, 2, 709−715 Probing the role of defects and disorder on optoelectronic properties requires sensitive spectroscopic probes. Defect transitions typically have much longer absorption lengths than interband transitions, therefore requiring analysis of optical absorption over several orders of magnitude below that at Eg. To this end, photothermal deflection spectroscopy (PDS) and Fourier transform photocurrent spectroscopy (FTPS) have been performed on isolated thin films, and subbandgap external quantum efficiency (EQE), transient photocapacitance, and transient photocurrent measurements have been applied to full photovoltaic devices. For CH3NH3PbI3 films, previous PDS and FTPS measurements have shown purely exponential absorptance behavior with steep slopes that correspond to low values of Eu (∼15 meV) and without the signatures of deep states within the bandgap for up to 4 orders of magnitude of response below Eg. 10 Similarly low values of Eu have been observed for CH3NH3PbBr3 films by PDS (∼25 meV) and devices by photocurrent spectroscopy (∼14 meV). Values on mixed bromide/iodide alloys CH3NH3Pb(I1−xBrx)3 have been more varied. Sadhanala et al. observed higher values of Eu at intermediate x with a peak of 90 meV for x = 0.8. In contrast, Hoke et al. observed low values of Eu = 12−17 meV over nearly the full range of x, with an outlier at x = 0.5. Although the methods described above have been successful at characterizing Eu in CH3NH3Pb(I1−xBrx) materials and solar cells, the signatures of electronically active defect states residing deep in the gap are rarely observed. In this regard, highly sensitive sub-bandgap EQE is a particularly powerful approach because it is capable of probing ultralow photocurrents due to sub-bandgap absorption that can be several orders of magnitude weaker than absorption at Eg. High sensitivity is enabled by using a high-throughput monochromator, a low-noise-current preamplifier, and careful filtering of above-bandgap stray light, allowing investigation of the sub-bandgap DOS (see the Supporting Information for experimental details). Indeed, this method has recently been applied to CH3NH3PbI3 solar cells to reveal a defect band at 1.34 eV with varying density that was linked to photovoltaic characteristics. Here, we present sub-bandgap EQE measurements of Au/ spiro-OMeTAD/CH3NH3Pb(I1−xBrx)3/TiO2/FTO/glass photovoltaic devices at room temperature and extending over the full halide composition space (0 ≤ x ≤ 1). In addition, we assess the impact of long-term aging of devices under atmospheric conditions, as monitored via formation of PbI2, on photoluminescence quantum yields (PLQYs) and Urbach energies. Samples were aged for up to 2300 h in a typical laboratory environment with a temperature of 22 ± 1 °C and a relative humidity of 50 ± 7%. For all compositions, we found Urbach energies in the range of 15−23 meV that remain approximately constant throughout the experiment. Moreover, we found direct evidence of sub-bandgap electronic states within the perovskite light absorbers. For pure CH3NH3PbI3, a single, relatively shallow state was observed. With increasing Br content, this state moves closer to the midgap while another shallow defect is revealed. If the densities of defect states near midgap cannot be controlled, then they will likely enhance Shockley−Read−Hall recombination in higher bandgap perovskites and potentially limit device performance. Our results provide important insight into the defect characteristics of the explored perovskite compositions, with implications for understanding disorderand defect-induced performance limits. CH3NH3Pb(I1−xBrx)3 films on FTO/TiO2 substrates were synthesized by a two-step, low-pressure, vapor-assisted, solution process (LP-VASP) that yields high material uniformity and quality, with previously reported solar cell power conversion efficiencies of up to 19%. In brief, the process starts with spin-coating and drying of the PbI2/PbBr2 precursor on the substrate in a N2 atmosphere for 15 min at 110 °C and is followed by annealing in CH3NH3I and CH3NH3Br vapor for 2 h at 120 °C (schematic inset of Figure 1a; details on device processing can be found in the Supporting Information and ref 17). This process allows access to the full halide composition space and results in the expected decrease of the lattice parameter with increasing Br content, as evaluated by the (110) X-ray diffraction (XRD) peak angle. In Figure 1a, the (110) peak angle is plotted versus the halide composition of the film, as measured by energy-dispersive X-ray spectroscopy (10 kV, FEI Quanta FEG 250), and is approximately linear over this small range of angles. The halide composition, x, in CH3NH3Pb(I1−xBrx)3 was related to the optical bandgap as described by the empirical equation Eg(x) = 1.598 + 0.36x + 0.34x and is displayed as the upper x-axis in Figure 1a. As mentioned above, sub-bandgap EQE measurements were Figure 1. (a) Change in the lattice parameter, as evaluated by the (110) XRD peak angle, plotted versus the halide composition, x, determined by EDX measurements. The methylammonium lead halide (CH3NH3Pb(I1−xBrx)3) perovskite films were prepared by the vapor-assisted solution process (schematic inset). The upper xaxis provides the bandgap corresponding to composition, calculated according to ref 15. (b) False colored cross-sectional scanning electron microscopy image of a typical planar Au/spiroOMeTAD/CH3NH3Pb(I1−xBrx)3/TiO2/FTO photovoltaic device fabricated for this study. ACS Energy Letters Letter DOI: 10.1021/acsenergylett.6b00727 ACS Energy Lett. 2017, 2, 709−715 710 performed on photovoltaic devices. We fabricated the full stack of Au/spiro-OMeTAD/CH3NH3Pb(I1−xBrx)3/TiO2/FTO/ glass with a planar device architecture, as illustrated in the cross-sectional scanning electron microscopy image in Figure 1b. The sub-bandgap EQE was measured using long-pass filters to limit above-bandgap stray light from a scanning monochromator to 1 part in 10 throughout most of the spectral range (more experimental details can be found in the Supporting Information and in ref 12). Given that lightinduced halide segregation has been reported to create Iand Br-rich domains, it is important to note that the monochromatic light intensity used during EQE measurements was <0.04 mW/cm above the bandgap of all samples examined. The light intensity varied over the spectral measurement and was below 0.4 mW/cm at 0.8 eV, 0.04 mW/cm at 1.6 eV, and 0.01 mW/cm at 2.3 eV. This is much lower than the 10−100 mW/cm used in the studies in which varying degrees of light-induced halide segregation were observed. The proposed mechanisms for halide segregation are predicated on the generation of mobile charges and ions; thus, low irradiances may limit halide segregation during the measurement. Furthermore, according to prior work, halide segregation in these alloys is fully reversible within a few minutes of relaxation in the dark. Because the EQE measurements reported here are performed at least 5 min after exposure to above-bandgap light more intense than that used for the measurement itself and the sub-bandgap portion requires hours of data collection, the role of halide segregation should be limited. Nonetheless, light-induced halide segregation in mixed I/Br samples cannot be fully excluded and may contribute, in parallel to defect responses, to sub-bandgap absorption. However, transitions in the samples with x = 0 and 1 and below 1.59 eV cannot be explained by the formation of Irich domains. Representative sub-bandgap EQE spectra of CH3NH3Pb(I1−xBrx)3 devices (with x = 0, 0.4, and 1) are shown in Figure 2a together with the underlying Gaussian defect distributions plotted as solid blue and magenta lines (more sub-bandgap EQE spectra are illustrated in Figure S1 as well as full EQE spectra on a linear scale in Figure S2, Supporting Information). They reveal photocurrent due to optical transitions in the bandgap that likely involve point defects within the perovskite thin films. Sub-bandgap defect-related transitions in the EQE spectra (EQEd) were fit using a function of the form σ ∝ + − ⎛ ⎝⎜ ⎞ ⎠⎟ E E EQE 1 erf 2 d d